35 lines
2.1 KiB
Python
Executable File
35 lines
2.1 KiB
Python
Executable File
from .cfg_optimization import cfg_skip
|
|
from .discrete_sampler import DiscreteSampling
|
|
from .flex_chunking import (broadcast_chunk_sizes, build_pyramid_partitions,
|
|
chunk_boundaries, chunk_ends_tensor,
|
|
chunk_sizes_to_block_kwargs, normalize_chunk_spec,
|
|
refine_partition, sample_flexible_chunks,
|
|
uniform_chunks, validate_nested_partitions)
|
|
from .fm_solvers import FlowDPMSolverMultistepScheduler
|
|
from .fm_solvers_minimax_h3 import MiniMaxH3Scheduler
|
|
from .fm_solvers_unipc import FlowUniPCMultistepScheduler
|
|
from .fp8_optimization import (autocast_model_forward,
|
|
convert_model_weight_to_float8,
|
|
convert_weight_dtype_wrapper,
|
|
replace_parameters_by_name)
|
|
from .fsdp_ema import FSDPEMA, LORAFSDPEMA
|
|
from .group_offload import (register_auto_device_hook,
|
|
safe_enable_group_offload,
|
|
safe_remove_group_offloading)
|
|
from .lora_utils import (LoadedTaomateH3LoRA, TaomateH3LoRACheckpointError,
|
|
canonical_taomate_h3_lora_targets, convert_peft_lora_to_kohya_lora,
|
|
convert_taomate_h3_adapter, create_network, load_taomate_h3_adapter,
|
|
merge_lora, official_to_kohya_lora_state_dict, unmerge_lora)
|
|
from .perf_metrics import install as install_perf_metrics
|
|
from .perf_metrics import install_training as install_perf_training
|
|
from .perf_metrics import instrument_pipeline
|
|
from .sd3_sde_with_logprob import sde_step_with_logprob
|
|
from .tqdm_bar import PauseAwareTqdm
|
|
from .trigflow_sampler import (RectifiedFlow_TrigFlowWrapper,
|
|
sample_trigflow_timesteps)
|
|
from .utils import (SegmentVideoSaver, StreamVideoSaver, calculate_dimensions,
|
|
filter_kwargs, get_autocast_dtype, get_image_latent,
|
|
get_image_to_video_latent, get_video_to_video_latent,
|
|
save_videos_grid, save_videos_with_audio_grid)
|
|
from .utils_yolo import ObjectDetector, ObjectInstanceDetector
|